Prospective buyers that Anthropic's AI buying agent hands to its sales team convert more than twice as often as leads from the old contact form, and deals close about five days faster, according to figures the company reported in October 2026.

Those are Anthropic's own numbers, not the result of an independent experiment. But the mechanism behind them is worth a close look for any revenue team weighing where AI agents belong in the funnel.

What changed in the inbound workflow?

Before, companies wanting to buy Claude filled out a contact form and sometimes waited days for an answer — even when their questions were straightforward. Buyers wanted to know the price, the minimum seat count, or whether Anthropic could meet a specific contract requirement.

The information existed. The bottleneck was capacity: Anthropic's sales team fielded tens of thousands of inbound requests per month and could not respond fast enough.

The company has since replaced much of that experience with an AI buying agent. The agent answers questions, works out what a buyer needs, recommends a plan, and can take the customer through checkout — end to end, without a human in the loop for routine purchases.

Why did conversion improve?

The agent did not lift results by removing salespeople from every purchase. It absorbed the repetitive questions and early qualification work that preceded any useful conversation.

By the time a representative got involved, two things had happened:

• The buyer understood the product better.
• The representative knew what the buyer actually needed.

That reframing matters for pipeline design. The measurable gains here are qualification quality and cycle compression at the top of the funnel — not raw inquiry throughput.

What should sales leaders take from it?

The more valuable ambition for a sales agent is not simply handling a larger volume of inquiries, but raising the quality of the conversations that reach humans. Anthropic's case shows the pattern: automate the pre-sale answers, qualify before handoff, and let reps start further down the funnel.

The usual caveats apply. The conversion and cycle-time gains are self-reported by a vendor that also builds the underlying model, with no published methodology or sample size. Teams replicating the approach should benchmark their own form-to-meeting, meeting-to-opportunity, and opportunity-to-close rates before and after any agent rollout.

Still, the direction of the result is consistent: buyers with pre-answered questions and pre-established fit convert faster. Anthropic has said it will keep publishing details of how its sales team rebuilt inbound with managed agents, which should give other operators more data to test the claim against their own pipelines.